Uveal melanomas are both sight and life threatening. Radiation therapy and enucleation are equally limited in preventing metastasis, and there is no effective treatment for metastatic uveal melanoma. There are no lymphatics in uveal melanomas: these tumors spread exclusively by a hematogenous route. The uveal melanoma microcirculation is heterogeneous, including incorporated pre-existing vessels, mosaic vessels, angiogenic vessels, and perfusable extra-vascular matrix patterns (EMPs) now regarded as a "fluid conducting meshwork". The presence of looping and interconnected EMPs in primary melanomas is strongly associated with death from metastases: the primary tumors of 88% of patients dying of metastatic uveal melanoma contain EMPs which are also present in all sites of metastatic uveal melanoma. EMPs are generated in vitro by aggressive uveal melanoma cells that are embedded in matrix in the absence of endothelial cells and fibroblasts; non-aggressive melanoma cells do not generate EMPs, regardless of matrix conditions, addition of soluble growth factors and hypoxia. EMPs connect to tumor blood vessels and conduct plasma in animal models and human tissues. Preliminary data indicate that the 3-dimensional distribution of perfused EMPs may provide a greater surface area exposure to plasma and therapeutic agents than vessels. We propose in the first specific aim to compare the distribution of plasma in EMPs with blood vessels in an orthotopic animal model and in human tissues using novel 3-dimensional visualization techniques. In the second aim, we take advantage of the fact that EMPs are generated by aggressive uveal melanoma cells only when these cells are embedded in thick matrix. By growing aggressive uveal melanoma cell lines under conditions permissive and not permissive of EMP formation, we can compare differential gene expression regulating EMP formation. By validating expression of these discriminating genes in human primary uveal melanoma tissues with EMPs from patients who have died of metastatic melanoma, and in metastatic uveal melanoma tissues, we can identify novel molecular targets for the treatment of aggressive primary uveal melanomas and their metastases.